WO2018227902A1 - 一种多通道回声消除电路、方法和智能设备 - Google Patents

一种多通道回声消除电路、方法和智能设备 Download PDF

Info

Publication number
WO2018227902A1
WO2018227902A1 PCT/CN2017/115229 CN2017115229W WO2018227902A1 WO 2018227902 A1 WO2018227902 A1 WO 2018227902A1 CN 2017115229 W CN2017115229 W CN 2017115229W WO 2018227902 A1 WO2018227902 A1 WO 2018227902A1
Authority
WO
WIPO (PCT)
Prior art keywords
output
input
channel
signal
echo cancellation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2017/115229
Other languages
English (en)
French (fr)
Inventor
朱剑
张向东
于振宇
罗志平
严栋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Goertek Inc
Original Assignee
Goertek Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Goertek Inc filed Critical Goertek Inc
Priority to US15/780,156 priority Critical patent/US10643634B2/en
Publication of WO2018227902A1 publication Critical patent/WO2018227902A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208—Noise filtering
    • G10L21/0216—Noise filtering characterised by the method used for estimating noise
    • G10L21/0232—Processing in the frequency domain
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; ELECTRIC HEARING AIDS; PUBLIC ADDRESS SYSTEMS
    • H04R3/00—Circuits for transducers
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04M—TELEPHONIC COMMUNICATION
    • H04M9/00—Arrangements for interconnection not involving centralised switching
    • H04M9/08—Two-way loud-speaking telephone systems with means for conditioning the signal, e.g. for suppressing echoes for one or both directions of traffic
    • H04M9/082—Two-way loud-speaking telephone systems with means for conditioning the signal, e.g. for suppressing echoes for one or both directions of traffic using echo cancellers
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208—Noise filtering
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; ELECTRIC HEARING AIDS; PUBLIC ADDRESS SYSTEMS
    • H04R3/00—Circuits for transducers
    • H04R3/02—Circuits for transducers for preventing acoustic reaction, i.e. acoustic oscillatory feedback
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; ELECTRIC HEARING AIDS; PUBLIC ADDRESS SYSTEMS
    • H04R5/00—Stereophonic arrangements
    • H04R5/02—Spatial or constructional arrangements of loudspeakers
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; ELECTRIC HEARING AIDS; PUBLIC ADDRESS SYSTEMS
    • H04R5/00—Stereophonic arrangements
    • H04R5/04—Circuit arrangements, e.g. for selective connection of amplifier inputs/outputs to loudspeakers, for loudspeaker detection, or for adaptation of settings to personal preferences or hearing impairments
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04S—STEREOPHONIC SYSTEMS 
    • H04S3/00—Systems employing more than two channels, e.g. quadraphonic
    • H04S3/008—Systems employing more than two channels, e.g. quadraphonic in which the audio signals are in digital form, i.e. employing more than two discrete digital channels
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208—Noise filtering
    • G10L2021/02082—Noise filtering the noise being echo, reverberation of the speech
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; ELECTRIC HEARING AIDS; PUBLIC ADDRESS SYSTEMS
    • H04R2227/00—Details of public address [PA] systems covered by H04R27/00 but not provided for in any of its subgroups
    • H04R2227/009—Signal processing in [PA] systems to enhance the speech intelligibility
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; ELECTRIC HEARING AIDS; PUBLIC ADDRESS SYSTEMS
    • H04R2430/00—Signal processing covered by H04R, not provided for in its groups
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; ELECTRIC HEARING AIDS; PUBLIC ADDRESS SYSTEMS
    • H04R27/00—Public address systems
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; ELECTRIC HEARING AIDS; PUBLIC ADDRESS SYSTEMS
    • H04R3/00—Circuits for transducers
    • H04R3/12—Circuits for transducers for distributing signals to two or more loudspeakers
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04S—STEREOPHONIC SYSTEMS 
    • H04S2400/00—Details of stereophonic systems covered by H04S but not provided for in its groups
    • H04S2400/01—Multi-channel, i.e. more than two input channels, sound reproduction with two speakers wherein the multi-channel information is substantially preserved

Definitions

  • the present invention relates to the field of audio signal processing technologies, and in particular, to a multi-channel echo cancellation circuit, method, and smart device.
  • the first problem encountered is the interference of external noise, especially the sound played by the product's own speaker. Because the speaker and the microphone are attached to the product, the distance in the space is very close, and the audio played by the speaker is easily transmitted back to the system through the microphone, so that the voice command spoken by the user is covered by the noise of the speaker, and the system cannot be given to the system. The command is sent and the system does not recognize the user voice command. This phenomenon is called echo interference. At present, echo interference has become the first problem to be solved in the design of voice interaction for intelligent hardware products.
  • Echo cancellation technology has emerged, but current echo cancellation technology generally only supports the elimination of one speaker, ie single channel noise. In fact, most products often design two channels or even more channels corresponding to two speakers, and single-channel noise elimination can no longer meet the needs of actual products.
  • the invention provides a multi-channel echo cancellation circuit, method and intelligent device for eliminating echo of multi-channel products and meeting actual needs.
  • a multi-channel echo cancellation circuit comprising: a processor, a signal extraction circuit and an isolation circuit corresponding to an audio channel of the speaker;
  • One end of the signal extraction circuit is connected to the audio channel of the corresponding speaker, and is used for extracting a part of the audio signal from the audio channel of the corresponding speaker as an echo cancellation reference signal;
  • the other end of the signal extraction circuit is connected to the isolation circuit
  • Each of the isolation circuits is connected to each other to form a noise channel, and the echo cancellation reference signal extracted by each signal extraction circuit forms a noise signal to the processor;
  • the processor subtracts the noise signal from the sound signal collected by the microphone according to the input noise signal and the acquired sound signal of the microphone to obtain a noise-reduced signal.
  • a multi-channel echo cancellation method comprising:
  • a part of the audio signal is extracted from the audio channel of each corresponding speaker as an echo cancellation reference signal by a signal extraction circuit, wherein the signal extraction circuit has the same number of audio channels as the speaker, and one end of the signal extraction circuit is connected with the audio channel of the corresponding speaker, and the signal is extracted. The other end of the circuit is connected to the isolation circuit;
  • Isolating circuits connected to the other end of the signal extraction circuit are connected to each other to form a noise channel, and the echo cancellation reference signal is formed into a noise signal through the noise channel to be output to the processor;
  • the processor acquires the sound signal collected by the microphone, and subtracts the noise signal from the sound signal collected by the microphone according to the input noise signal and the acquired sound signal of the microphone to obtain the noise-reduced signal.
  • a smart device comprising a multi-channel echo cancellation circuit as in the previous aspect.
  • the multi-channel echo cancellation circuit of the embodiment of the invention supports the echo cancellation of the product of the multi-channel multi-speaker, and separately extracts the audio of the plurality of channels by the signal extraction circuit corresponding to the number of audio channels of the speaker Information, a noise channel formed by interconnected isolation circuits is coupled into a noise signal and transmitted to the processor, so that it can support multi-channel echo cancellation, improve product echo cancellation performance, and thereby improve product human-computer interaction. accuracy.
  • the isolation circuits are connected to each other to form a noise channel, the echo cancellation reference signals extracted from the audio channels can be aggregated into one noise signal and provided to the processor, so that the processor only needs to have an audio interface for receiving noise.
  • the requirements on the processor are reduced, and the cost is saved.
  • the design of the isolation circuit avoids crosstalk between the multi-speakers, the sound quality of the multi-speaker output is improved, and the market competitiveness of the product is improved.
  • FIG. 1 is a block diagram of a multi-channel echo cancellation circuit in accordance with some embodiments of the present invention.
  • FIG. 2 is a circuit diagram of a multi-channel echo cancellation circuit in accordance with some embodiments of the present invention.
  • FIG. 3 is a block diagram of a multi-channel echo cancellation circuit according to further embodiments of the present invention.
  • FIG. 4 is a schematic flow chart of a multi-channel echo cancellation method according to some embodiments of the present invention.
  • FIG. 5 is a structural block diagram of a smart device according to some embodiments of the present invention.
  • Echo cancellation is to first acquire audio and digital processing from the audio channel of the speaker, obtain noise samples, and collect voice signals and noise from the microphone.
  • the audio processing is performed by DSP (Digital Signal Processing) processor, and the microphone is collected into the audio signal.
  • the noise is removed to obtain a clean user voice signal, which is then transmitted to the relevant post-stage system for speech recognition. It can be seen that echo cancellation is crucial for speech recognition and voice interaction of products. Prior art echo cancellation techniques are generally only applied to mono.
  • this scheme can be applied to multiple channels, but multiple channels are directly connected through signal lines, there is not enough isolation between them, and there is sound crosstalk between the speakers.
  • multi-channel direct connection means that the ground lines of multiple audios are connected together, which will generate additional common ground noise. After these common ground noises are transmitted to the DSP, the noise reduction performance of the echo cancellation is significantly reduced, and unwanted noise is also generated, which adversely affects the sound quality of the speaker.
  • embodiments of the present invention provide a multi-channel echo cancellation circuit that requires only one noise audio channel, is low in cost, and can remove various unwanted audio noise interferences.
  • the design concept of the technical solution of the present invention is to provide an isolation circuit, for example, a transformer is used in the isolation circuit.
  • the working principle of the transformer is inductive coupling, which is commonly used to transform voltage or change the impedance of the load.
  • the multi-channel echo cancellation circuit of the embodiment includes: a processor 103, a signal extraction circuit 101 corresponding to the audio channel of the speaker, and an isolation circuit 102. ;
  • One end of the signal extraction circuit 101 is connected to the audio channel of the corresponding speaker for extracting a part of the audio signal from the audio channel of the corresponding speaker as an echo cancellation reference signal;
  • the other end of the signal extraction circuit 101 is connected to the isolation circuit 102;
  • Each of the isolation circuits 102 is connected to each other to form a noise channel, and the echo cancellation extracted by each signal extraction circuit 101 is eliminated.
  • the reference signal forming a noise signal output to the processor 103;
  • the processor 103 subtracts the noise signal from the sound signal collected by the microphone according to the input noise signal and the acquired sound signal collected by the microphone to obtain a noise-reduced signal.
  • the echo cancellation circuit of the embodiment of the present invention extracts a part of the signal from the audio signal of the speaker as an echo cancellation reference signal by using a signal extraction circuit connected to the audio channel of each speaker, and then passes through the signal extraction circuit.
  • the connected isolation circuit outputs to the processor. Since the isolation circuits are connected, the echo cancellation reference signal can form a noise signal, and the noise signal is sent to the processor through a noise channel for noise reduction processing. In this way, the processor only needs one audio port to receive a noise signal output by the isolation circuit, and does not need to set multiple audio ports, which reduces the requirement on the processor and saves cost.
  • the signal extraction circuits of the audio channels are isolated, and the signal extraction does not affect each other.
  • a noise channel is formed through the connected isolation circuit, and a noise path formed by the noise channel is formed.
  • the signal is output to the processor, avoiding the influence of crosstalk when the multi-sound channel is directly connected for signal extraction, and avoiding the generation of additional common ground noise, thereby improving the noise reduction performance of the echo cancellation and ensuring the sound quality of the speaker.
  • FIG. 1 is an example in which two channels, two signal extraction circuits corresponding to two channels, and two isolation circuits are taken as an example, but embodiments of the present invention are not limited thereto.
  • the isolation circuit in this embodiment includes: a first stage transformer (such as the transformer N1 illustrated in FIG. 2) and an output matching network 22 connected to the output end of the first stage transformer, and the signal extraction circuit is an input matching.
  • the network 21 the audio signal corresponding to the speaker (for example, the left speaker) is input to the matching network 21 to obtain an echo cancellation reference signal, and the echo cancellation reference signal is input to the corresponding first stage transformer, and each output matching network 22 is connected to each other to form a noise.
  • the channel 23 outputs a noise signal formed by the echo cancellation reference signal to the processor.
  • the first-stage transformer of this embodiment selects a low-frequency transformer, and the range of the operating frequency of the transformer is consistent with the frequency range of the sound signal output to the speaker.
  • the operating frequency range of the transformer is set in the range of 10 to 20000 Hz, which is consistent with the frequency range of the sound signal of the speaker to realize conversion and reverse isolation of the audio signal.
  • the input matching network and the output matching network are composed of passive components including resistors and capacitors.
  • the input matching network 21 includes: a first input resistor (the resistor R11 illustrated in FIG. 2), a first input capacitor (capacitor C11 illustrated in FIG. 2), and a second input resistor (FIG. 2).
  • the resistor R12 the resistor R12
  • the input of the first-stage transformer (the transformer N1 as shown in FIG. 2) is connected in series with the first input capacitor C11, Connected to one end of the second input resistor R12 and one end of the first input resistor R11, the other end of the second input resistor R12 is connected to the first ground end of the first stage transformer N1, the first ground end and the speaker (such as the left speaker)
  • the ground line (ie, GND3 illustrated in FIG. 2) is connected, and the other end of the first input resistor R11 is connected to the audio signal line of the speaker. That is, the input end of the first-stage transformer N1 is connected in series with the first input capacitor C11, and then connected in parallel with the second input resistor R12, and finally the first input resistor R11 is connected in series.
  • the output matching network 22 includes a first output resistor (resistor R31 illustrated in FIG. 2), a first output capacitor (capacitor C32 illustrated in FIG. 2), and a second output resistor (resistor R32 illustrated in FIG. 2).
  • the output of the first-stage transformer (such as the transformer N1 illustrated in FIG. 2) is connected in series with the first output resistor R31, and is connected to one end of the first output capacitor C32 and one end of the second output resistor R32, and the second output resistor R32 The other end is connected to the second ground end of the first stage transformer N1, and the second ground end is connected to the ground line of the processor (ie, GND2 illustrated in FIG. 2).
  • the plurality of GND2s illustrated in FIG. 2 are echoes of the embodiment. Eliminate the common ground of the circuit.
  • the output terminal of the first-stage transformer N1 is connected in series with the first output resistor R31, and then connected in parallel with the second output resistor R32, and finally the first output capacitor C32 is connected in series.
  • the ground line of the audio signal line of the embodiment is completely isolated from the processor ground line, thereby avoiding hum noise caused by the ground line of the speaker and the ground of the processor sharing one ground.
  • the other end of the first output capacitor C32 is connected to the noise channel 23, and the noise channel 23 is connected to the audio input terminal (the audio input terminal, that is, the audio port) of the processor.
  • the first input capacitor C11 and the first output capacitor C32 are DC blocking capacitors.
  • a DC blocking capacitor with a capacitance of 4.7 ⁇ F can be selected.
  • the DC blocking capacitor is a capacitor that blocks the DC component of the signal and allows the AC component to pass smoothly to the subsequent circuit.
  • the input matching network is formed by the first input resistor, the first input capacitor and the second input resistor, and the partial signal is extracted from the sound channel of the corresponding speaker as the echo canceling reference signal.
  • the function of the input matching network of this embodiment is to match the output impedance of the speaker with the input impedance of the transformer, that is, to ensure that the power signal of the working frequency of the transformer is extracted from the high-powered speaker signal, and to some extent suppress the harmonics outside the working frequency. Wave components and interference.
  • the role of the output matching network is to ensure that the output impedance of the transformer matches the input impedance of the processor, reducing line reflections and noise interference.
  • the device selection of the input matching network is determined by the following formula:
  • R 12 is the resistance of the second input resistor
  • Z 1 is the input impedance of the first stage transformer
  • R 11 is the resistance of the first input resistor
  • a 1 is the first stage transformer (such as the transformer N1 in FIG. 2)
  • the preset transformation ratio; the transformer's transformation ratio refers to the ratio of the input voltage to the output voltage of the transformer.
  • the audio signal extraction function is realized by selecting an appropriate transformer input voltage and output voltage ratio, that is, extracting a small power signal from the high-power speaker audio channel for echo cancellation. Since the reverse attenuation of the transformer is as high as 20 dB, it is equivalent to only 1/100 of the signal can be transmitted backwards, thus avoiding crosstalk between the left and right channels and improving the performance of echo cancellation.
  • the echo cancellation circuit of the embodiment After extracting the low-power echo cancellation reference signal through the input matching network connected to the sound channels of the respective speakers, the echo cancellation circuit of the embodiment passes through the transformer and the output matching network connected to the transformer to realize the impedance of the matching transformer and the processor.
  • the purpose of the impedance so as to avoid the impedance mismatch between the transformer and the processor, resulting in reverse signal transmission, affecting the problem of echo cancellation performance.
  • the device selection of the output matching network is determined by the following conditions:
  • R 31 is the resistance of the first output resistor
  • K 1 is the output impedance of the first stage transformer
  • R 32 is the resistance of the second output resistor
  • G is the input impedance of the audio port of the processor.
  • the signal processing process of the echo cancellation circuit shown in FIG. 2 is: the input matching network is connected to the left and right sound channels, and the sound signals of the left and right channels are acquired, and then input to the transformers N1 and N2 for inductive coupling and power conversion. After the audio output signals of the transformers N1 and N2 pass through the two output matching networks respectively, a noise signal is formed, and then input into an echo canceling device such as a DSP for processing to obtain a signal output after noise reduction.
  • an echo canceling device such as a DSP for processing to obtain a signal output after noise reduction.
  • the circuit design including the matching network and the transformer ensures that the signal input to the DSP contains left and right audio information of the same power, and echo cancellation can be better achieved.
  • the isolation circuit includes a first-stage transformer.
  • the isolation circuit may further include: a second-stage transformer, specifically, an output end of each output matching network and a second The input terminals of the stage transformer are connected, a noise channel is formed by the second stage transformer, and the echo cancellation reference signal forms a noise signal to the processor.
  • the solid black point of the output matching network corresponding to the left and right channels in FIG. 2 can be replaced by a transformer, that is, a second-stage transformer is set, so that the multi-path echo is set by setting the second-stage transformer.
  • the structure of the echo canceling circuit when two channels are set in the product is described above. Similarly, when the product includes more than two channels, the structure of the echo canceling circuit is similar. When more echo cancellation is required, only a corresponding number of input matching networks need to be connected, and the matching network and the transformer can be output.
  • the first channel is connected to the first input matching network, and the first input matching network and the first channel A transformer is connected, and the first transformer is connected to the first output network.
  • the second channel is coupled to the second input matching network, the second input matching network is coupled to the second transformer, and the second transformer is coupled to the second output network.
  • the third channel is connected to the third input matching network, the third input matching network is connected to the third transformer, and the third transformer is connected to the third output network.
  • the first output network, the second output network and the third output network are connected to construct a noise channel, and a formed noise signal is output.
  • more audio channels can be connected to the transformer after passing through the input matching network, and then the output of the transformer is also passed through the output matching network and then connected to a noise channel, thereby passing a path through a noise channel.
  • the noise signal is sent to the processor for processing.
  • FIG. 4 is a schematic flowchart of a multi-channel echo cancellation method according to another embodiment of the present invention.
  • the multi-channel echo cancellation method includes the following steps:
  • Step S401 extracting a part of the audio signal from the audio channel of each corresponding speaker as an echo cancellation reference signal by using a signal extraction circuit, wherein the number of audio channels of the signal extraction circuit and the speaker is the same, and the signal is extracted.
  • One end of the road is connected to the audio channel of the corresponding speaker, and the other end of the signal extraction circuit is connected to the isolation circuit;
  • Step S402 connecting the isolation circuits connected to the other end of the signal extraction circuit to each other to form a noise channel, and outputting the echo cancellation reference signal to form a noise signal to the processor through the noise channel;
  • Step S403 the processor acquires the sound signal collected by the microphone, and subtracts the noise signal from the sound signal collected by the microphone according to the input noise signal and the acquired sound signal of the microphone, to obtain the noise-reduced signal.
  • the isolation circuits connected to the other end of the signal extraction circuit are connected to each other to form a noise channel.
  • the first stage transformer is selected, and the output matching network is connected to the output end of the first stage transformer.
  • the method includes: constructing an input matching network by using a passive component, and using an input matching network as a signal extraction circuit, wherein the passive component includes a resistor and a capacitor;
  • Constructing the input matching network by using passive components includes: selecting a first input resistor, a first input capacitor and a second input resistor, connecting the input end of the first stage transformer to the first input capacitor, and then the second input resistor Parallel, and finally connect the first input resistors in series to construct an input matching network.
  • the input end of the first-stage transformer is connected in series with the first input capacitor, and is connected to one end of the second input resistor and one end of the first input resistor, and the other end of the second input resistor is connected to the first ground of the first-stage transformer.
  • the terminal is connected, the first ground is connected to the ground of the speaker, and the other end of the first input resistor is connected to the audio signal line of the speaker.
  • the method includes: constructing an output matching network by using passive components, the passive component includes a resistor and a capacitor; and constructing the output matching network by using the passive component includes: selecting the first output resistor, the first output capacitor, and the first The output resistor combines the output of the first stage transformer with the first output resistor, and then parallels with the second output resistor, and finally connects the first output capacitor in series to construct an output matching network.
  • the output end of the first-stage transformer is connected in series with the first output resistor, and is connected to one end of the first output capacitor and one end of the second output resistor, and the other end of the second output resistor is connected to the second ground of the first-stage transformer.
  • the terminal is connected, and the second ground is connected to the ground of the processor.
  • the other end of the first output capacitor is connected to the noise channel, and the noise channel is connected to the audio input end of the processor; the first input capacitor and the first output capacitor are DC blocking capacitors, first The operating frequency range of the stage transformer is consistent with the frequency range of the sound signal output to the speaker.
  • the first stage transformer is selected as a low frequency transformer, and the operating frequency of the first stage transformer ranges from 10 to 20000 Hz, and the capacitance of the first input capacitor and the first output capacitor is selected to be 4.7. ⁇ F.
  • Forming a noise channel includes: selecting a first-stage transformer, connecting an output matching network at an output end of the first-stage transformer, connecting an output matching network to an input end of a second-stage transformer, and outputting the output ends of the second-stage transformers A noise channel is formed after the connection.
  • the isolation circuits connected to the other end of the signal extraction circuit are connected to each other to form a noise channel.
  • the method further includes: selecting a second-stage transformer, and matching the output ends of the output matching networks with the second The input terminals of the stage transformer are connected, an isolation circuit is constructed, and a noise channel is formed by the second stage transformer.
  • the device selection of the input matching network satisfies the following conditions:
  • R 12 is the resistance of the second input resistor
  • Z 1 is the input impedance of the first stage transformer
  • R 11 is the resistance of the first input resistor
  • a 1 is the preset transformation ratio of the first stage transformer.
  • the device selection of the output matching network satisfies the following conditions:
  • R 31 is the resistance of the first output resistor
  • K 1 is the output impedance of the first stage transformer
  • R 32 is the resistance of the second output resistor
  • G is the input impedance of the audio port of the processor.
  • a smart device in the embodiment of the present invention, is provided, and the smart device 50 includes a multi-channel echo cancellation circuit 501, which is the multi-channel echo cancellation circuit in the foregoing embodiment.
  • the smart device of the embodiment of the invention improves the echo cancellation performance, thereby improving the user's voice interaction experience and improving the market competitiveness of the smart device.
  • the echo cancellation circuit and method of the embodiment of the present invention eliminates the reference signal by echo extracted from each speaker channel, and then forms a noise signal to be output to the processor through the noise channel to support multi-channel echo. Elimination of processing reduces the need for processors, and the processor only needs one noise channel to save costs. In addition, there is sufficient isolation between the sound channels of the individual speakers to avoid crosstalk problems. Moreover, the speaker is sufficiently isolated from the GND network of the processor to avoid common ground noise and improve echo cancellation performance.

Landscapes

  • Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Quality & Reliability (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Otolaryngology (AREA)
  • Circuit For Audible Band Transducer (AREA)
  • Telephone Function (AREA)
  • Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)

Abstract

一种多通道回声消除电路、方法和智能设备。多通道回声消除电路包括信号提取电路(101)，一端与对应扬声器的音频通道连接，从对应扬声器的音频通道中提取部分音频信号作为回声消除参考信号；信号提取电路(101)的另一端连接隔离电路(102)；各隔离电路(102)相互连接后形成一噪声通道，将各信号提取电路(101)提取的回声消除参考信号形成一路噪声信号输出给处理器(103)，处理器(103)根据输入的噪声信号以及麦克风采集的声音信号，将麦克风采集的声音信号减去噪声信号得到降噪后的信号。该多通道回声消除电路和方法，支持多声道回声消除，提高了回声消除性能，改善了多声道音频的输出音质，包括多通道回声消除电路的智能设备，用户体验更佳。

Description

一种多通道回声消除电路、方法和智能设备 技术领域
本发明涉及音频信号处理技术领域，具体涉及一种多通道回声消除电路、方法和智能设备。
背景技术
随着电子信息技术和声学技术的发展以及机器人等智能硬件产品的发展，基于语音识别技术的人机交互技术越来越多的在手机、机器人、智能音响、智能电视等等产品上应用。
语音识别的人机交互，在实际应用中，遇到的第一个问题就是外部噪声的干扰，特别是产品自身扬声器播放的声音。因为扬声器跟麦克风都附着在产品上，空间上距离很近，扬声器播放的音频很轻易就通过麦克风传回到系统当中去，这样用户说的语音指令就被扬声器的噪声覆盖掉了，无法给系统发送指令，系统也无法识别用户语音指令。这一现象叫做回声干扰。目前回声干扰已经成为智能硬件产品的语音交互设计中首先要解决的问题。
回声消除技术应运而生，但是目前的回声消除技术一般只支持一个扬声器、即单路噪声的消除。而实际上大部分的产品往往设计对应两个扬声器的两个声道甚至更多声道，单路噪声消除已经无法满足实际产品的需求。
发明内容
本发明提供了一种多通道回声消除电路、方法和智能设备，以对多声道产品的回声进行消除，满足实际需求。
根据本发明的一个方面，提供了一种多通道回声消除电路，包括：处理器、与扬声器的音频通道对应数量的信号提取电路和隔离电路；
信号提取电路一端与对应扬声器的音频通道连接，用于从对应扬声器的音频通道中提取部分音频信号作为回声消除参考信号；
信号提取电路的另一端连接隔离电路；
各隔离电路相互连接后形成一噪声通道，将各信号提取电路提取的回声消除参考信号形成一路噪声信号输出给处理器；
处理器，根据输入的噪声信号以及获取的麦克风采集的声音信号，将麦克风采集的声音信号减去噪声信号，得到降噪后的信号。
根据本发明的另一个方面，提供了一种多通道回声消除方法，包括：
通过信号提取电路从各对应扬声器的音频通道中提取部分音频信号作为回声消除参考信号，其中，信号提取电路与扬声器的音频通道数量相同，信号提取电路的一端与对应扬声器的音频通道连接，信号提取电路的另一端连接隔离电路；
将与信号提取电路的另一端连接的隔离电路相互连接，形成一噪声通道，通过噪声通道将回声消除参考信号形成一路噪声信号输出给处理器；
处理器获取麦克风采集的声音信号，并根据输入的噪声信号以及获取的麦克风采集的声音信号，将麦克风采集的声音信号减去噪声信号，得到降噪后的信号。
根据本发明的又一个方面，提供了一种智能设备，该智能设备包括如前述一个方面的多通道回声消除电路。
本发明的有益效果是：本发明实施例的多通道回声消除电路，支持多声道多扬声器的产品的回声消除，通过与扬声器的音频通道对应数量的信号提取电路分别提取多个声道的音频信息，经相互连接的隔离电路形成的一噪声通道耦合成一路噪声信号后传送到处理器，如此，既能够支持多声道的回声消除，提高产品的回声消除性能，进而提高产品的人机交互准确性。又由于各隔离电路相互连接后形成一噪声通道，使得从各音频通道中提取的回声消除参考信号能汇聚成一路噪声信号，提供给处理器，使得处理器只需要具备一个接收噪声的音频接口即可，降低了对处理器的要求，节省成本；同时由于隔离电路的设计避免了多扬声器之间的串音，也改善了多扬声器输出的音质，提升了产品的市场竞争力。
附图说明
图1是本发明一些实施例的多通道回声消除电路的框图；
图2是本发明一些实施例的多通道回声消除电路的电路图；
图3是本发明另一些实施例的多通道回声消除电路的框图；
图4是本发明一些实施例的多通道回声消除方法的流程示意图；
图5是本发明一些实施例的智能设备的结构框图。
具体实施方式
回声消除是先从扬声器音频通道进行音频采集与数字处理，得到噪声样本，并从麦克风采集语音信号与噪声，通过DSP(Digital Signal Processing数字相关处理)处理器进行音频处理，把麦克风采集到音频信号中的噪声去除，从而得到干净的用户语音信号，再传给相关的后级系统进行语音识别等。可知，回声消除对产品的语音识别和语音交互至关重要。现有技术的回声消除技术一般只应用于单声道。
有一种回声消除方案：这种方案可以应用于多声道，但是多声道直接通过信号线连接，相互之间没有足够的隔离度，扬声器之间会有声音串扰。
例如当左声道以最大音量播放，右声道静默时，由于串音的影响，从右声道的扬声器中也能听到左声道播放的音乐。同时，多声道直接连接也就意味着多个音频的地线连接在一起，会产生额外的共地噪声。这些共地噪声输送到DSP后，导致回声消除的降噪性能明显下降，同时也会出现不需要的杂音，对扬声器的音质产生不良影响。
还有一种回声消除方案：每一个扬声器的音频通道，都单独的提取噪声信号后发送给DSP进行回声消除。但是这种方式要求DSP需要有相应的多个音频端口，而且DSP的回声消除要进行多次的降噪处理，成本较高。
对此，本发明实施例提供了一种多通道回声消除电路，只需要一条噪声音频通道，成本低，并且可以去除各种不需要的音频噪声干扰。本发明技术方案的设计构思在于设置隔离电路，例如在隔离电路中采用变压器，变压器的工作原理是电感耦合，常用于变换电压或变换负载的阻抗。通过将变压器的工作频率范围设置为10～20000Hz，与扬声器的声音信号的频率范围重合，能够实现音频信号的转换和反向隔离，提高降噪效果。
参见图1，为本发明一些实施例的多通道回声消除电路的框图，本实施例的多通道回声消除电路包括：处理器103、与扬声器的音频通道对应数量的信号提取电路101和隔离电路102；
信号提取电路101一端与对应扬声器的音频通道连接，用于从对应扬声器的音频通道中提取部分音频信号作为回声消除参考信号；
信号提取电路101的另一端连接隔离电路102；
各隔离电路102相互连接后形成一噪声通道，将各信号提取电路101提取的回声消 除参考信号形成一路噪声信号输出给处理器103；
处理器103，根据输入的噪声信号以及获取的麦克风采集的声音信号，将麦克风采集的声音信号减去噪声信号，得到降噪后的信号。
由图1所示可知，本发明实施例的回声消除电路，利用与各扬声器的音频通道连接的信号提取电路，从扬声器的音频信号中提取部分信号作为回声消除参考信号，然后经过与信号提取电路连接的隔离电路输出给处理器，由于各个隔离电路相连接，从而可以将回声消除参考信号形成一路噪声信号，通过一个噪声通道将这一路噪声信号发送给处理器供其进行降噪处理。这样处理器只需要一个音频端口来接收隔离电路输出的一路噪声信号，不需要设置多个音频端口，降低了对处理器的要求，节省了成本。另外，本实施例中各音频通道的信号提取电路是隔离的，信号提取相互不影响，在提取回声消除参考信号后再经过相连接的隔离电路形成一个噪声通道，经噪声通道将形成的一路噪声信号输出给处理器，避免了多声音通道直接连接进行信号提取时串音的影响，也避免了额外的共地噪声的产生，从而提高了回声消除的降噪性能，并保证了扬声器的音质。
需要说明的是，图1中是以两声道，以及与两个声道对应的两个信号提取电路和两个隔离电路为例进行的说明，但本发明的实施例不限于此。
如图2所示，本实施例中隔离电路包括：第一级变压器(如图2中示意的变压器N1)和与第一级变压器的输出端连接的输出匹配网络22，信号提取电路为输入匹配网络21，对应扬声器(例如，左扬声器)的音频信号通过输入匹配网络21后得到回声消除参考信号，回声消除参考信号输入到对应的第一级变压器，各输出匹配网络22相互连接后形成一噪声通道23，将回声消除参考信号形成的一路噪声信号输出给处理器。
需要强调的是，本实施例的第一级变压器选取低频变压器，变压器的工作频率的范围与输出到扬声器的声音信号的频率范围一致。例如，将变压器的工作频率范围设置在10～20000Hz范围内，与扬声器的声音信号的频率范围一致，以实现音频信号的转换和反向隔离。
输入匹配网络和输出匹配网络通过无源元件组成，无源元件包括电阻、电容。参见图2，本实施例中，输入匹配网络21包括：第一输入电阻(图2中示意的电阻R11)，第一输入电容(图2中示意的电容C11)和第二输入电阻(图2中示意的电阻R12)，
第一级变压器(如图2中示意的变压器N1)的输入端与第一输入电容C11串联后， 与第二输入电阻R12的一端以及第一输入电阻R11的一端连接，第二输入电阻R12的另一端与第一级变压器N1的第一接地端连接，第一接地端与扬声器(如左扬声器)的地线(即图2中示意的GND3)连接，第一输入电阻R11的另一端连接扬声器的音频信号线。即第一级变压器N1的输入端与第一输入电容C11串联后，再与第二输入电阻R12并联，最后再串联第一输入电阻R11。
输出匹配网络22包括：第一输出电阻(图2中示意的电阻R31)，第一输出电容(图2中示意的电容C32)和第二输出电阻(图2中示意的电阻R32)。
第一级变压器(如图2中示意的变压器N1)的输出端与第一输出电阻R31串联后，与第一输出电容C32的一端以及第二输出电阻R32的一端连接，第二输出电阻R32的另一端与第一级变压器N1的第二接地端连接，第二接地端与处理器的地线(即图2中示意的GND2)连接，图2中示意的多处GND2是本实施例的回声消除电路的公共地。即，第一级变压器N1的输出端与第一输出电阻R31串联后，再与第二输出电阻R32并联，最后再串联第一输出电容C32。由图2可知，本实施例的音频信号线的地线和处理器地线完全隔离，避免了由于扬声器的地线和处理器的地线共用一个地导致的交流声干扰。
第一输出电容C32的另一端与噪声通道23连接，噪声通道23与处理器的音频输入端(音频输入端即音频端口)连接。这里的第一输入电容C11和第一输出电容C32为隔直电容，一般地，可以选用容值为4.7μF的隔直电容。隔直电容是指将信号中直流成分阻断，而让交流成分顺利传递到后级电路的电容。
上述是以左声道为例对电路结构进行的说明，由于图2中右声道的电路结构和左声道的相同，因此右声道的电路结构不再重复赘述。
由图2所述可知，本实施例中通过第一输入电阻、第一输入电容和第二输入电阻构成输入匹配网络，实现从对应的扬声器的声音通道中提取部分信号作为回声消除参考信号的目的。本实施例的输入匹配网络的作用是实现扬声器的输出阻抗与变压器的输入阻抗相匹配，即，保证从大功率的扬声器信号中提取变压器工作频率的功率信号，一定程度上抑制工作频率以外的谐波分量和干扰。输出匹配网络的作用是保证变压器的输出阻抗与处理器的输入阻抗相匹配，减小线路反射和噪声干扰。
需要说明的是，由于扬声器的功率一般较大，在从扬声器的音频通道中声音信号中 提取部分信号时，需要选择合适的变压比来确定提取的信号的功率。本实施例中通过如下公式确定输入匹配网络的器件选型：
Figure PCTCN2017115229-appb-000001
其中，R12为第二输入电阻的阻值，Z1为第一级变压器的输入阻抗，R11为第一输入电阻的阻值，A1为第一级变压器(如图2中的变压器N1)的预设变压比值；变压器的变压比值是指变压器的输入电压和输出电压的比值。
实际应用中，通过选择合适的变压器输入电压和输出电压比值，实现音频信号的提取功能，即，从大功率的扬声器音频通道中提取一个小功率信号供回声消除使用。由于变压器的反向衰减高达20dB，相当于只有1/100的信号可以逆向传送，这样就避免了左右声道之间的串音问题，提高了回声消除的性能。
右声道连接的输入匹配网络的器件选型参见前述公式，这里不再赘述。
在通过与各扬声器的声音通道连接的输入匹配网络提取出小功率的回声消除参考信号之后，本实施例回声消除电路经过变压器以及与变压器连接的输出匹配网络，实现匹配变压器的阻抗和处理器的阻抗的目的，从而避免变压器和处理器的阻抗不匹配，导致信号逆向传送，影响回声消除性能的问题。
为了达到阻抗匹配，本实施例中通过如下条件确定输出匹配网络的器件选型：
Figure PCTCN2017115229-appb-000002
其中，R31为第一输出电阻的阻值，K1为第一级变压器的输出阻抗，R32为第二输出电阻的阻值，G为处理器音频端口的输入阻抗。
如图2所示的回声消除电路的信号处理过程是：输入匹配网络连接左右声音通道，获取左、右声道的声音信号之后，输入到变压器N1和N2进行电感耦合和功率变换。变压器N1和N2的音频输出信号分别经过两个输出匹配网络之后，再形成一路噪声信号，然后输入到回声消除器件如DSP中进行处理，得到降噪后的信号输出。
由此，通过包含匹配网络和变压器的电路设计保证输入到DSP中的信号包含相同功率的左、右音频信息，可以更好的实现回声消除。
上述实施例中以隔离电路包含第一级变压器进行了说明，在本发明的另一些实施例中，隔离电路还可以包括：第二级变压器，具体的，各输出匹配网络的输出端与第二级变压器的输入端连接，通过第二级变压器形成一噪声通道，将回声消除参考信号形成一路噪声信号输出给处理器。
结合图2，图2中对应左、右声道的输出匹配网络相连接的实心黑点处可以用一个变压器替换，即设置第二级变压器，这样，通过设置第二级变压器，将多路回声消除参考信号合成一路噪声信号，借助变压器的隔离特性，实现各声音通道之间的隔离，保证只有音频信号通过，除了声音信号之外的其它信号被滤除，以减少噪声干扰，提高回声消除性能。
以上介绍了产品中设置两个声道时的回声消除电路的结构，同样的，当产品中包括两个以上的声道时，回声消除电路的结构相似。当需要进行更多声道的回声消除时，只需要接入对应数量的输入匹配网络，输出匹配网络和变压器即可。如图3所示，当产品中有三个声音通道，即第一声道、第二声道和第三声道时，第一声道与第一输入匹配网络连接，第一输入匹配网络与第一变压器连接，第一变压器与第一输出网络连接。第二声道与第二输入匹配网络连接，第二输入匹配网络与第二变压器连接，第二变压器与第二输出网络连接。第三声道与第三输入匹配网络连接，第三输入匹配网络与第三变压器连接，第三变压器与第三输出网络连接。第一输出网络，第二输出网络和第三输出网络相连接构建一个噪声通道，将形成的一路噪声信号输出。
同样的，其它实施例中可以把更多的音频通道在通过输入匹配网络后接入变压器，然后把变压器的输出也经过输出匹配网络后接入到一个噪声通道中，从而通过一个噪声通道将一路噪声信号发送给处理器供其处理。
由于多个声音通道的回声消除电路的工作过程和前述图2所示的电路的工作过程相同，因此这里不再重复说明。
与前述多通道回声消除电路相对应的，本发明实施例还提供了一种多通道回声消除方法，图4是本发明另一些实施例的多通道回声消除方法的流程示意图；如图4所示，多通道回声消除方法包括如下步骤：
步骤S401，通过信号提取电路从各对应扬声器的音频通道中提取部分音频信号作为回声消除参考信号，其中，信号提取电路与扬声器的音频通道数量相同，信号提取电 路的一端与对应扬声器的音频通道连接，信号提取电路的另一端连接隔离电路；
步骤S402，将与信号提取电路的另一端连接的隔离电路相互连接，形成一噪声通道，通过噪声通道将回声消除参考信号形成一路噪声信号输出给处理器；
步骤S403，处理器获取麦克风采集的声音信号，并根据输入的噪声信号以及获取的麦克风采集的声音信号，将麦克风采集的声音信号减去噪声信号，得到降噪后的信号。
在本发明的一些实施例中，将与信号提取电路的另一端连接的隔离电路相互连接，形成一噪声通道包括：选取第一级变压器，在第一级变压器的输出端连接输出匹配网络，构建隔离电路；将各输出匹配网络相互连接后形成一噪声通道。
在本发明的一些实施例中，包括：利用无源元件构建输入匹配网络，将输入匹配网络作为信号提取电路，其中，无源元件包括电阻、电容；
利用无源元件构建输入匹配网络包括：选取第一输入电阻，第一输入电容和第二输入电阻，将第一级变压器的输入端与所述第一输入电容串联后，再与第二输入电阻并联，最后再串联第一输入电阻，构建输入匹配网络。具体地，第一级变压器的输入端与第一输入电容串联后，与第二输入电阻的一端以及第一输入电阻的一端连接，第二输入电阻的另一端与第一级变压器的第一接地端连接，第一接地端与扬声器的地线连接，第一输入电阻的另一端连接扬声器的音频信号线。
在本发明的一些实施例中，包括：利用无源元件构建输出匹配网络，无源元件包括电阻、电容；利用无源元件构建输出匹配网络包括：选取第一输出电阻，第一输出电容和第二输出电阻，将第一级变压器的输出端与第一输出电阻串联后，再与第二输出电阻并联，最后再串联第一输出电容，构建输出匹配网络。具体地，第一级变压器的输出端与第一输出电阻串联后，与第一输出电容的一端以及第二输出电阻的一端连接，第二输出电阻的另一端与第一级变压器的第二接地端连接，第二接地端与处理器的地线连接。
在本发明的一些实施例中，第一输出电容的另一端与噪声通道连接，噪声通道与所述处理器的音频输入端连接；第一输入电容和第一输出电容为隔直电容，第一级变压器的工作频率的范围与输出到扬声器的声音信号的频率范围一致。
在本发明的一些实施例中，选取第一级变压器为低频变压器，第一级变压器的工作频率范围为10～20000Hz，选取所述第一输入电容和所述第一输出电容的容值为4.7μF。
在本发明的一些实施例中，将与信号提取电路的另一端连接的隔离电路相互连接， 形成一噪声通道包括：选取第一级变压器，在第一级变压器的输出端连接输出匹配网络，将输出匹配网络与一个第二级变压器的输入端连接，将各第二级变压器的输出端相连接后形成一噪声通道。
在本发明的一些实施例中，将与信号提取电路的另一端连接的隔离电路相互连接，形成一噪声通道还包括：选取第二级变压器，将各所述输出匹配网络的输出端与第二级变压器的输入端连接，构建隔离电路，通过第二级变压器形成一噪声通道。
在本发明的一些实施例中，将输入匹配网络的器件选型满足如下条件：
Figure PCTCN2017115229-appb-000003
其中，R12为第二输入电阻的阻值，Z1为第一级变压器的输入阻抗，R11为第一输入电阻的阻值，A1为第一级变压器的预设变压比值。
在本发明的一些实施例中，将所述输出匹配网络的器件选型满足如下条件：
Figure PCTCN2017115229-appb-000004
其中，R31为第一输出电阻的阻值，K1为第一级变压器的输出阻抗，R32为第二输出电阻的阻值，G为处理器音频端口的输入阻抗。
另外，本发明实施例中提供了一种智能设备，该智能设备50包括多通道回声消除电路501，该多通道回声消除电路501即为前述实施例中的多通道回声消除电路。通过采用本发明实施例的多通道回声消除电路，使得本发明实施例的智能设备，提高了回声消除性能，进而改善了用户的语音交互体验，提升了智能设备的市场竞争力。
综上所述，本发明实施例的回声消除电路和方法，通过分别从各扬声器声道提取的回声消除参考信号，然后形成一个噪声信号经噪声声道输出给处理器，支持多声道的回声消除处理，降低了对于处理器的要求，处理器只需要一个噪声通道，节省成本。另外，各个扬声器的声音通道之间保持足够的隔离，避免了串音问题。而且，扬声器与处理器的GND地网络保持足够的隔离，避免产生共地噪声，提高了回声消除性能。
以上所述，仅为本发明的具体实施方式，在本发明的上述教导下，本领域技术人员可以在上述实施例的基础上进行其他的改进或变形。本领域技术人员应该明白，上述的 具体描述只是更好的解释本发明的目的，本发明的保护范围以权利要求的保护范围为准。

Claims (15)

  1. 一种多通道回声消除电路，包括处理器、与扬声器的音频通道对应数量的信号提取电路和隔离电路；
    所述信号提取电路一端与对应扬声器的音频通道连接，用于从对应扬声器的音频通道中提取部分音频信号作为回声消除参考信号；
    所述信号提取电路的另一端连接隔离电路；
    各所述隔离电路相互连接后形成一噪声通道，将各所述信号提取电路提取的回声消除参考信号形成一路噪声信号输出给所述处理器；
    所述处理器，根据输入的所述噪声信号以及获取的麦克风采集的声音信号，将麦克风采集的声音信号减去所述噪声信号，得到降噪后的信号。
  2. 根据权利要求1所述的多通道回声消除电路，其中，所述隔离电路包括：第一级变压器和与所述第一级变压器的输出端连接的输出匹配网络，
    所述信号提取电路为输入匹配网络，对应扬声器的音频信号通过所述输入匹配网络后得到回声消除参考信号，回声消除参考信号输入到对应的第一级变压器，各所述输出匹配网络相互连接后形成一噪声通道，将所述回声消除参考信号形成一路噪声信号输出给所述处理器。
  3. 根据权利要求2所述的多通道回声消除电路，其中，所述隔离电路还包括：第二级变压器，
    各所述输出匹配网络的输出端与所述第二级变压器的输入端连接，通过所述第二级变压器形成一噪声通道，将所述回声消除参考信号形成一路噪声信号输出给所述处理器。
  4. 根据权利要求2所述的多通道回声消除电路，其中，所述输入匹配网络包括：第一输入电阻，第一输入电容和第二输入电阻，
    第一级变压器的输入端与所述第一输入电容串联后，与所述第二输入电阻的一端以及所述第一输入电阻的一端连接，所述第二输入电阻的另一端与所述第一级变压器的第一接地端连接，所述第一接地端与扬声器的地线连接，所述第一输入电阻的另一端连接所述扬声器的音频信号线；
    所述输出匹配网络包括：第一输出电阻，第一输出电容和第二输出电阻，
    第一级变压器的输出端与所述第一输出电阻串联后，与所述第一输出电容的一端以及所述第二输出电阻的一端连接，所述第二输出电阻的另一端与所述第一级变压器的第二接地端连接，所述第二接地端与处理器的地线连接，所述第一输出电容的另一端与所述噪声通道连接，所述噪声通道与所述处理器的音频输入端连接。
  5. 根据权利要求4所述的多通道回声消除电路，其中，所述输入匹配网络的器件选型满足如下条件：
    Figure PCTCN2017115229-appb-100001
    其中，R12为第二输入电阻的阻值，Z1为第一级变压器的输入阻抗，R11为第一输入电阻的阻值，A1为第一级变压器的预设变压比值；
    输出匹配网络的器件选型满足如下条件：
    Figure PCTCN2017115229-appb-100002
    其中，R31为第一输出电阻的阻值，K1为第一级变压器的输出阻抗，R32为第二输出电阻的阻值，G为处理器音频端口的输入阻抗。
  6. 根据权利要求4所述的多通道回声消除电路，其中，
    所述第一输入电容和所述第一输出电容为隔直电容；
    所述第一级变压器的工作频率范围与输出到扬声器的声音信号的频率范围一致。
  7. 根据权利要求6所述的多通道回声消除电路，其中，
    所述第一级变压器为低频变压器，所述第一级变压器的工作频率范围为10～20000Hz；
    所述第一输入电容和所述第一输出电容的容值为4.7μF。
  8. 一种多通道回声消除方法，该方法包括：
    通过信号提取电路从各对应扬声器的音频通道中提取部分音频信号作为回声消除参考信号，其中，所述信号提取电路与扬声器的音频通道数量相同，信号 提取电路的一端与对应扬声器的音频通道连接，信号提取电路的另一端连接隔离电路；
    将与信号提取电路的另一端连接的隔离电路相互连接，形成一噪声通道，通过所述噪声通道将所述回声消除参考信号形成一路噪声信号输出给处理器；
    处理器获取麦克风采集的声音信号，并根据所述输入的噪声信号以及获取的麦克风采集的声音信号，将麦克风采集的声音信号减去所述噪声信号，得到降噪后的信号。
  9. 根据权利要求8所述的多通道回声消除方法，其中，所述将与信号提取电路的另一端连接的隔离电路相互连接，形成一噪声通道包括：
    选取第一级变压器，在所述第一级变压器的输出端连接输出匹配网络，构建隔离电路，将各所述输出匹配网络相互连接后形成一噪声通道。
  10. 根据权利要求9所述的多通道回声消除方法，其中，包括：利用无源元件构建输入匹配网络和输出匹配网络，将所述输入匹配网络作为所述信号提取电路，其中，所述无源元件包括电阻、电容；
    所述利用无源元件构建输入匹配网络包括：选取第一输入电阻，第一输入电容和第二输入电阻，
    将第一级变压器的输入端与所述第一输入电容串联后，再与第二输入电阻并联，最后再串联第一输入电阻，构建输入匹配网络，其中，第一级变压器的输入端与所述第一输入电容串联后，与所述第二输入电阻的一端以及所述第一输入电阻的一端连接，所述第二输入电阻的另一端与所述第一级变压器的第一接地端连接，所述第一接地端与扬声器的地线连接，所述第一输入电阻的另一端连接所述扬声器的音频信号线。
    所述利用无源元件构建输出匹配网络包括：选取第一输出电阻，第一输出电容和第二输出电阻，
    将第一级变压器的输出端与第一输出电阻串联后，再与第二输出电阻并联，最后再串联第一输出电容，构建输出匹配网络，其中，第一级变压器的输出端与所述第一输出电阻串联后，与所述第一输出电容的一端以及所述第二输出电阻的一端连接，所述第二输出电阻的另一端与所述第一级变压器的第二接地端连接， 所述第二接地端与处理器的地线连接，所述第一输出电容的另一端与所述噪声通道连接，所述噪声通道与所述处理器的音频输入端连接。
  11. 根据权利要求10所述的多通道回声消除方法，其中，
    所述第一输入电容和所述第一输出电容为隔直电容，所述第一级变压器的工作频率的范围与输出到扬声器的声音信号的频率范围一致。
  12. 根据权利要求9所述的多通道回声消除方法，其中，所述将与信号提取电路的另一端连接的隔离电路相互连接，形成一噪声通道还包括：
    选取第二级变压器，将各所述输出匹配网络的输出端与所述第二级变压器的输入端连接，构建隔离电路，通过所述第二级变压器形成一噪声通道。
  13. 根据权利要求10所述的多通道回声消除方法，其中，将所述输入匹配网络的器件选型满足如下条件：
    Figure PCTCN2017115229-appb-100003
    其中，R12为第二输入电阻的阻值，Z1为第一级变压器的输入阻抗，R11为第一输入电阻的阻值，A1为第一级变压器的预设变压比值。
    将所述输出匹配网络的器件选型满足如下条件：
    Figure PCTCN2017115229-appb-100004
    其中，R31为第一输出电阻的阻值，K1为第一级变压器的输出阻抗，R32为第二输出电阻的阻值，G为处理器音频端口的输入阻抗。
  14. 根据权利要求13所述的多通道回声消除方法，其中，选取所述第一级变压器为低频变压器，所述第一级变压器的工作频率范围为10～20000Hz；
    选取所述第一输入电容和所述第一输出电容的容值为4.7μF。
  15. 一种智能设备，包括：如权利要求1-7任一所述的多通道回声消除电路。
PCT/CN2017/115229 2017-06-15 2017-12-08 一种多通道回声消除电路、方法和智能设备 Ceased WO2018227902A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US15/780,156 US10643634B2 (en) 2017-06-15 2017-12-08 Multichannel echo cancellation circuit and method and smart device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710452304.5A CN107105366B (zh) 2017-06-15 2017-06-15 一种多通道回声消除电路、方法和智能设备
CN201710452304.5 2017-06-15

Publications (1)

Publication Number Publication Date
WO2018227902A1 true WO2018227902A1 (zh) 2018-12-20

Family

ID=59660359

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/115229 Ceased WO2018227902A1 (zh) 2017-06-15 2017-12-08 一种多通道回声消除电路、方法和智能设备

Country Status (3)

Country Link
US (1) US10643634B2 (zh)
CN (1) CN107105366B (zh)
WO (1) WO2018227902A1 (zh)

Families Citing this family (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107105366B (zh) * 2017-06-15 2022-09-23 歌尔股份有限公司 一种多通道回声消除电路、方法和智能设备
US10511909B2 (en) * 2017-11-29 2019-12-17 Boomcloud 360, Inc. Crosstalk cancellation for opposite-facing transaural loudspeaker systems
CN108335701B (zh) * 2018-01-24 2021-04-13 青岛海信移动通信技术股份有限公司 一种进行声音降噪的方法及设备
CN110956973A (zh) * 2018-09-27 2020-04-03 深圳市冠旭电子股份有限公司 一种回声消除方法、装置及智能终端
WO2020063798A1 (zh) * 2018-09-27 2020-04-02 深圳市冠旭电子股份有限公司 一种回声消除方法、装置及智能音箱
CN111356058B (zh) * 2018-12-20 2021-08-20 深圳市冠旭电子股份有限公司 一种回声消除方法、装置及智能音箱
CN111261179A (zh) * 2018-11-30 2020-06-09 阿里巴巴集团控股有限公司 回声消除方法及装置和智能设备
CN109495082A (zh) * 2018-12-29 2019-03-19 合肥惠科金扬科技有限公司 一种音频传输的多级匹配电路、方法及终端设备
CN110086434A (zh) * 2019-02-28 2019-08-02 厦门优迅高速芯片有限公司 一种提升跨阻放大电路中rssi脚抗噪能力的电路
CN110310654A (zh) * 2019-07-26 2019-10-08 歌尔科技有限公司 回声消除方法及装置、电子设备、可读存储介质
CN111182118B (zh) * 2020-01-03 2021-05-25 维沃移动通信有限公司 一种音量调节方法及电子设备
CN111756943B (zh) * 2020-06-05 2021-04-20 中原工学院 一种增广复值成比例子带自适应立体声学回声消除方法
CN111726464B (zh) * 2020-06-29 2021-04-20 珠海全志科技股份有限公司 一种多通道回声滤除方法、滤除装置和可读存储介质
CN113316047B (zh) * 2021-04-16 2023-04-14 杭州涂鸦信息技术有限公司 一种拾音设备
CN115696140B (zh) * 2022-12-05 2023-05-26 长沙东玛克信息科技有限公司 一种教室音频多通道回声消除方法
CN116261081B (zh) * 2023-05-16 2023-08-01 合肥联宝信息技术有限公司 一种耳机电路及其控制方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101562669A (zh) * 2009-03-11 2009-10-21 屈国良 自适应全双工全频段回声消除的方法
US20130177162A1 (en) * 2012-01-09 2013-07-11 Via Telecom, Inc. Novel pre-process (amplitude distortion) and post-process (phase synchronization) for linear aec system
CN104464739A (zh) * 2013-09-18 2015-03-25 华为技术有限公司 音频信号处理方法及装置、差分波束形成方法及装置
CN107105366A (zh) * 2017-06-15 2017-08-29 歌尔股份有限公司 一种多通道回声消除电路、方法和智能设备

Family Cites Families (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3060265A (en) * 1960-01-26 1962-10-23 Itt Conference call circuit
IT1246839B (it) * 1990-07-13 1994-11-28 Flaminio Frassinetti Apparecchiatura per la miscelazione, a separazione di banda, di segnali elettrici.
FR2762467B1 (fr) * 1997-04-16 1999-07-02 France Telecom Procede d'annulation d'echo acoustique multi-voies et annuleur d'echo acoustique multi-voies
DE69903012T2 (de) * 1998-04-08 2003-04-30 British Telecommunications P.L.C., London Echokompensation
US7110531B2 (en) * 2000-02-17 2006-09-19 Analog Devices, Inc. Isolation system with analog communication across an isolation barrier
US20020054685A1 (en) * 2000-11-09 2002-05-09 Carlos Avendano System for suppressing acoustic echoes and interferences in multi-channel audio systems
BRPI0704910E2 (pt) * 2007-09-26 2010-03-16 Fernando Eid Pires aperfeiçoamento introduzido em equipamento de áudio
CN102572644B (zh) * 2010-12-29 2015-07-15 上海博泰悦臻电子设备制造有限公司 麦克风装置
WO2012094528A1 (en) * 2011-01-05 2012-07-12 Conexant Systems, Inc., A Delaware Corporation Systems and methods for stereo echo cancellation
US8804977B2 (en) * 2011-03-18 2014-08-12 Dolby Laboratories Licensing Corporation Nonlinear reference signal processing for echo suppression
US20140079232A1 (en) * 2011-05-19 2014-03-20 Nec Corporation Audio processing device, audio processing method, and recording medium recording audio processing program
KR101306868B1 (ko) * 2011-12-21 2013-09-10 황정진 미확인 시스템 모델링 방법 및 이를 이용한 하울링 제거 오디오 시스템
US9445196B2 (en) * 2013-07-24 2016-09-13 Mh Acoustics Llc Inter-channel coherence reduction for stereophonic and multichannel acoustic echo cancellation
US9659555B1 (en) * 2016-02-09 2017-05-23 Amazon Technologies, Inc. Multichannel acoustic echo cancellation
CN106548783B (zh) * 2016-12-09 2020-07-14 西安Tcl软件开发有限公司 语音增强方法、装置及智能音箱、智能电视

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101562669A (zh) * 2009-03-11 2009-10-21 屈国良 自适应全双工全频段回声消除的方法
US20130177162A1 (en) * 2012-01-09 2013-07-11 Via Telecom, Inc. Novel pre-process (amplitude distortion) and post-process (phase synchronization) for linear aec system
CN104464739A (zh) * 2013-09-18 2015-03-25 华为技术有限公司 音频信号处理方法及装置、差分波束形成方法及装置
CN107105366A (zh) * 2017-06-15 2017-08-29 歌尔股份有限公司 一种多通道回声消除电路、方法和智能设备

Also Published As

Publication number Publication date
CN107105366A (zh) 2017-08-29
US10643634B2 (en) 2020-05-05
CN107105366B (zh) 2022-09-23
US20190362733A1 (en) 2019-11-28

Similar Documents

Publication Publication Date Title
WO2018227902A1 (zh) 一种多通道回声消除电路、方法和智能设备
CN104429100B (zh) 用于环绕声回声降低的系统及方法
US8731207B2 (en) Apparatus and method for computing control information for an echo suppression filter and apparatus and method for computing a delay value
TWI682672B (zh) 減少殘餘回聲的回聲消除系統和方法
US20210112157A1 (en) Method, apparatus, and computer-readable media utilizing residual echo estimate information to derive secondary echo reduction parameters
CA2593183C (en) Partitioned fast convolution in the time and frequency domain
CN103945076B (zh) 用于移动设备的实时音频回声和背景噪声的减小
CN105812598A (zh) 一种降低回声的方法及装置
WO2011044395A1 (en) An input signal mismatch compensation system
CN111356058B (zh) 一种回声消除方法、装置及智能音箱
CN105959874A (zh) 移动终端及其降低音频噪声的方法
TW202331701A (zh) 雙麥克風陣列回音消除方法、雙麥克風陣列回音消除裝置、電子設備、及非揮發性電腦可讀儲存媒體
US9491306B2 (en) Signal processing control in an audio device
CN111524532A (zh) 回声抑制方法、装置、设备及存储介质
CN106161820B (zh) 一种用于立体声声学回声抵消的通道间去相关方法
CN108834026A (zh) 一种基于ne5532芯片的音调调控电路
CN110876106A (zh) 电子设备、降噪方法、计算机系统和介质
CN101140760A (zh) 一种声音信号采集处理系统及方法
CN207022188U (zh) 一种多通道回声消除电路和智能设备
CN202059560U (zh) 麦克风装置
CN212992396U (zh) 一种回声消除电路、装置及视频系统
CN204906663U (zh) 麦克风装置及音响系统
TWI656525B (zh) 高保真語音裝置
US20130230183A1 (en) Echo cancellation
RU216993U1 (ru) Устройство для адаптивной фильтрации речи в системах аудио-конференц-связи

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17913686

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17913686

Country of ref document: EP

Kind code of ref document: A1